Edit: nowadays the same GPU is never going above 60C. What caused the extremely high temps was probably a combination of trapped air, solved by carefully moving and holding the radiator and GPU around (NOT SHAKING), and poorly applied thermal paste. I had a strange ticking noise that I mentioned in this thread and another that eventually completely went away after re-applying the thermal paste by finger in a thin layer. This also caused the temps to drop significantly. I recently swapped the stock radiator fan for a Phanteks T30 which helps a lot with getting air through the radiator at max RPM. Average temps are now between 40-50C at ~80% load and 50-60C at 90-100% load.
 
Bought it in 2016. Highest it reached back then was about 50-55C when overclocked with the values I still use today. I used the pre-applied thermal paste and hadn't changed it till last year when my temps started to reach 85+C in the summer. The temps were gradually increasing 2 years after I got the GPU before it reached that point. Idle temps are between 32-34C. Were below 30C when I first had the GPU.

The thermal paste I used to replace the old one was Arctic Silver 5. It was really old thermal paste, probably like 7-10 years old. It kind of worked because it got my temps down to 70C but only temporarily. It didn't take long for it to reach 80C again under full load. I replaced it once more with the same paste before I used Arctic MX-4 which I recently put on.

The issue I'm now having is quite weird. No matter how much thermal paste I use or how I apply it, I can't get it to not overheat. It'll reach up to 84C now under full load. I've applied the paste using the star method with more than enough thermal paste (probably too much since I used pretty thick lines), which is an X and a + so it covers every bit of the GPU die.

I've searched pretty heavily on what kind of issues others were having when reaching temps as high as I am and most people had a faulty water pump. I checked one of the hoses of the water pump and it's warm when I touch it under full load. The other one is cold. The air coming out of the radiator is warm as well but not as warm as I'd expect it to be at 84C though. The amount of air being pushed out seems to be rather weak too compared to when I first had the card. I noticed hot air comes out after the GPU reaches 70C.

Some people in other threads speculated on air bubbles getting stuck and suggested to shake the pump around a bit. I tried but to no avail. I can hear the water pump buzzing which means that it's working. One of the hoses is vibrating a little too.

The radiator is mounted to the back of my case blowing air out. Before it is a CPU cooler that blows air through it. I've checked the air blowing out of the CPU cooler and it's not hot so that's not the reason for my temps. Recently cleaned my case too which was pretty dusty but it didn't help with GPU temps.

I've tried putting all case fans and the GPU's shroud fan on max speeds but it barely helps. Maybe gets the temp down by 1-2C.

One other thing: ever since replacing the thermal paste on my GPU (year ago) I've been getting a weird ticking noise coming from it under >80% load, no matter how much I clean the GPU die and around it before applying thermal paste. My guess is that I keep using too much thermal paste when I'm applying but is this harmful? I didn't have the noise before I replaced the thermal paste for the first time.

I'm at a point where I'm considering buying a new card out of desperation but the card's performance still holds up today and hasn't disappointed me. Does anyone know if there's a way to get the pump working properly again?

MrModZix
Bought it in 2016. Highest it reached back then was about 50-55C when overclocked with the values I still use today. I used the pre-applied thermal paste and hadn't changed it till last year when my temps started to reach 85+C in the summer. The temps were gradually increasing 2 years after I got the GPU before it reached that point. Idle temps are between 32-34C. Were below 30C when I first had the GPU.

The thermal paste I used to replace the old one was Arctic Silver 5. It was really old thermal paste, probably like 7-10 years old. It kind of worked because it got my temps down to 70C but only temporarily. It didn't take long for it to reach 80C again under full load. I replaced it once more with the same paste before I used Arctic MX-4 which I recently put on.

The issue I'm now having is quite weird. No matter how much thermal paste I use or how I apply it, I can't get it to not overheat. It'll reach up to 84C now under full load. I've applied the paste using the star method with more than enough thermal paste (probably too much since I used pretty thick lines), which is an X and a + so it covers every bit of the GPU die.

I've searched pretty heavily on what kind of issues others were having when reaching temps as high as I am and most people had a faulty water pump. I checked one of the hoses of the water pump and it's warm when I touch it under full load. The other one is cold. The air coming out of the radiator is warm as well but not as warm as I'd expect it to be at 84C though. The amount of air being pushed out seems to be rather weak too compared to when I first had the card. I noticed hot air comes out after the GPU reaches 70C.

Some people in other threads speculated on air bubbles getting stuck and suggested to shake the pump around a bit. I tried but to no avail. I can hear the water pump buzzing which means that it's working. One of the hoses is vibrating a little too.

The radiator is mounted to the back of my case blowing air out. Before it is a CPU cooler that blows air through it. I've checked the air blowing out of the CPU cooler and it's not hot so that's not the reason for my temps. Recently cleaned my case too which was pretty dusty but it didn't help with GPU temps.

I've tried putting all case fans and the GPU's shroud fan on max speeds but it barely helps. Maybe gets the temp down by 1-2C.

One other thing: ever since replacing the thermal paste on my GPU (year ago) I've been getting a weird ticking noise coming from it under >80% load, no matter how much I clean the GPU die and around it before applying thermal paste. My guess is that I keep using too much thermal paste when I'm applying but is this harmful? I didn't have the noise before I replaced the thermal paste for the first time.

I'm at a point where I'm considering buying a new card out of desperation but the card's performance still holds up today and hasn't disappointed me. Does anyone know if there's a way to get the pump working properly again?

84C is kinda toasty with an AIO.  Personally I'd redo the thermal paste again, with Kryonaut TIM. GPU's need a bit more paste than your typical CPU, but you should not need an amount that it's squeezing out around the die area, even though that shouldn't cause any issues. A pea sized amount should be enough. When you remove the cooler are there ANY voids in the paste or does it look like you have full coverage?
I personally use two fans in push/pull exhaust on the radiator connected to a motherboard header. At 90% you can't hear either one,  couple of Scythe AP-14's. Gentle Typhoons.

Thanks for your reply.

I actually did some research about MX-4 vs Kryonaut and read that they only differ by about a few degrees C. That's why I just went with MX-4. It cost me half of what Kryonaut would cost me.

I read that when applying thermal paste on a GPU, the entire die should be covered because the GPU doesn't have an IHS like a CPU. That's why I kind of went overboard a little this time but I'm pretty sure the whole die is covered since I used as much as last time. When I reapplied recently the coverage was really good.

MrModZix
Thanks for your reply.

I actually did some research about MX-4 vs Kryonaut and read that they only differ by about a few degrees C. That's why I just went with MX-4. It cost me half of what Kryonaut would cost me.

I read that when applying thermal paste on a GPU, the entire die should be covered because the GPU doesn't have an IHS like a CPU. That's why I kind of went overboard a little this time but I'm pretty sure the whole die is covered since I used as much as last time. When I reapplied recently the coverage was really good.

Well yes, the entire die must be covered. When you removed the heat sink, does it appear that the TIM is squeezed out between the die and the cooler into a very thin even layer? If the heatsink is not flat and flush with the die, this can cause higher temps. How about the rad fan? Is it operating at the correct speed? If possible I'd try a better fan on the rad to see if that helps. The pump is a constant speed 12v unit so it either runs or it does not.

Sounds like the pump is probably degrading/failing..Aftermarket AIOs for graphics cards are available, such as something from NZXT/Kraken and others you could install.
 
Or, convert it to air cooling..Lots of aftermarket kits around for that too.
 
 
Some of that stuff ain't cheap, so I guess it depends on how important that card is to you vs buying another one.

I know you said you tinkered by shaking, which I personally never recommend as it can create/move more air bubbles inside the unit. I  have an old Corsair H60 that was super temperamental, and finally by running the system up-side down, and at some odd angle, all the air bubbles worked there way out of wherever they were wreaking havok, and life has been good since on that system(3+ years running). That's on an old Phenom II X4 940, which was pegging 70C at basically idle and shutting down system! Since fixing air bubble issue, even under complete stress that thing doesn't get above 52c. While it's possible your pump is going out, or getting weak, your RPM's in monitoring software such as Precision or Afterburner WOULD reflect it if so. There is plenty of really  great articles about thermal paste application, and at the end of the day, unless you're really globbing it on...........that is NOT your issue. Bad contact from the waterblock to the GPU itself could also be the issue, especially if the PCB has warped from excessive heat.

HeavyHemi
Well yes, the entire die must be covered. When you removed the heat sink, does it appear that the TIM is squeezed out between the die and the cooler into a very thin even layer? If the heatsink is not flat and flush with the die, this can cause higher temps. How about the rad fan? Is it operating at the correct speed? If possible I'd try a better fan on the rad to see if that helps. The pump is a constant speed 12v unit so it either runs or it does not.

You previously said a pea sized should be enough but I know from applying paste with that method on my CPU that it leaves a little of the corners uncovered. I thought you meant to say that leaving some bits uncovered wouldn't matter. That's why I mentioned it.

The paste is squeezed out but I'm not sure if it's very thin. I assume it is because I tightened the screws for the heatsink pretty tightly and used the cross tightening method.

The rad fan is spinning but I don't know at which RPM it's spinning. EVGA Precision only shows the shroud fan's RPM. Is it possible for the rad fan to degrade over time?

Your removed the radiator fans & vacuumed the radiator ... when you cleaned your PC ?
 
- heatsink needs to be kept parallel & properly tightened .... this maximizes the contact between heatsink & GPU die
 
- 1 hose warmer than the other .. liquid is circulating
 
- over time the coolant level could evaporate .. its happened - or - some sort of sediment has formed & impacting the flow
 
As long as your not using any type of Conductive TIM .... too much is not an issue

joeymir
I know you said you tinkered by shaking, which I personally never recommend as it can create/move more air bubbles inside the unit. I  have an old Corsair H60 that was super temperamental, and finally by running the system up-side down, and at some odd angle, all the air bubbles worked there way out of wherever they were wreaking havok, and life has been good since on that system(3+ years running). That's on an old Phenom II X4 940, which was pegging 70C at basically idle and shutting down system! Since fixing air bubble issue, even under complete stress that thing doesn't get above 52c. While it's possible your pump is going out, or getting weak, your RPM's in monitoring software such as Precision or Afterburner WOULD reflect it if so. There is plenty of really  great articles about thermal paste application, and at the end of the day, unless you're really globbing it on...........that is NOT your issue. Bad contact from the waterblock to the GPU itself could also be the issue, especially if the PCB has warped from excessive heat.

That's interesting. The overheating began before I shook it so I don't think that's really the cause.

I'm kind of thinking it's the rad fan since there's hot air coming out of the rad but it's not a lot. Almost like the rad can't dissipate enough heat.

Cool GTX
Your removed the radiator fans & vacuumed the radiator ... when you cleaned your PC ?
 
- heatsink needs to be kept parallel & properly tightened .... this maximizes the contact between heatsink & GPU die
 
- 1 hose warmer than the other .. liquid is circulating
 
- over time the coolant level could evaporate .. its happened - or - some sort of sediment has formed & impacting the flow
 
As long as your not using any type of Conductive TIM .... too much is not an issue

Yes, I removed the rad fan then cleaned the rad's and fan's dust.

I applied pressure with my thumb on the water pump and index finger on the backplate when mounting the pump back on.

What gets me is this: the "away from heat" hose gets warm really quickly whenever I put the GPU under stress. The air coming out of the radiator though starts getting hot only until the GPU reaches 70C. It really makes me think it's the rad fan at this point.

I thought the liquid couldn't evaporate since it's in a closed loop.

The TIM I used was Arctic MX-4.